In http://mail.python.org/pipermail/python-dev/2008-October/082994.html Martin v. Löwis wrote:
So 2.6.0 will contain a lot of tests that have never been tested in a wide variety of systems. Some are incorrect, and get fixed in 2.6.1, and stay fixed afterwards. This is completely different from somebody introducing a new test in 2.6.4. It means that there are more failures in a maintenance release, not less as in the first case.
If 2.6.1 has some (possibly accidental, but exposed to the users) behavior that is not a clear bug, it should be kept through 2.6.x. You may well want to change it in 2.7, but not in 2.6.4. Adding a test to 2.6.2 ensures that the behavior will not silently disappear because of an unrelated bugfix in 2.6.3.